From Napster to Now — How We Lost the Thread Part - 2
The rise of streaming, the false promise of democratization, and the quiet erasure of music’s meaning
Originally posted on Medium.
I was reading a book, MP3: The Meaning of a Format¹ by Jonathan Sterne (who tragically passed last month), which covers the fascinating 100-year history that led to the development of the MPEG Layer 3 codec, which became known as the MP3.
Towards the end, Sterne observes:
But because of the recording industry’s absence in the formative stages of the online music economy, music itself was demonetized.
This statement struck me as a pivotal insight. It reframes the familiar story of Napster and file-sharing not as a subversive blow against the titans of industry but as a historical rupture in music’s economic infrastructure.
Music wasn’t liberated by the MP3 or the networks it enabled through peer-to-peer platforms running on dial-up and early broadband. Instead, music as an object was stripped of its commodity status without being reimagined symbolically or structurally. The collapse of price came without a redefinition of value.
What followed in the streaming era was not a restoration. It was a strategic recapture that rebranded demonetization as democratization of access and, in doing so, erased the responsibilities we once held toward music and the musicians who create it. As Jonathan Sterne suggests, we have lost not just an economic model but also a relational one. What was framed as a right to music was never accompanied by a language of responsibility to music or the act of listening itself. When the CD economy crumbled, our response wasn’t thoughtful reinvention, it was technical implementation.
This piece follows and extends that thread. It explores how the demonetization of music was rebranded as democratization, how streaming consolidated rather than disrupted power, and why we are now entering a new dangerous phase shaped by generative AI and machinic logic whose ultimate telos is to obsolesce the body and transmute the production of art into pure artifice.
In Hirschmanian terms, the era of voice is over. The time for exit has arrived².
Sometime around 1991, I was really into the band Queensrÿche. That album Operation: Mindcrime was on repeat in my Walkman. A friend invited me to the mall with his parents, and while they wandered around, I went straight to the big record store. I looked for Queensrÿche in the cassette section and came up empty. There was a massive section for Queen, a band I remember hearing on MTV, but I did not know much about it at the time. In retrospect, this is kind of unbelievable, but that says more about pre-internet information asymmetries than anything else. I took a chance and bought their most recent release, Innuendo.
When I got home, I popped the tape into my tape deck and hated it. I was horrified. Not only did I hate it, but I’d just spent $12 on it. Plug that into the BLS inflation calculator and it comes out to $28.47 today. For that same amount today, I can get two and a half months of Spotify Premium and with that, I don’t just get Innuendo, I get access to basically the entire compendium of recorded music in human history.
A few months later I was up way past my approved bedtime watching MTV’s 120 Minutes when a song by a new band premiered — Smells Like Teen Spirit. I still remember the exact position I was in when Dave Grohl played his intro fill, and the video panned from a bunch of kids in the bleachers to his shaking head, and I never thought about that ‘wasted’ investment again.
While I was milling around in malls, taking chances on records by bands I didn’t know, a group of engineers were finalizing the specifications for what would become MPEG-1 Audio Layer III. What we now call the MP3 was formally released in 1992 as part of a suite of digital codecs, with the file extension .mp3 arriving a few years later. But it wasn’t until the mid-1990s that it broke out into the world.
In 1995, a hacker named SoloH came across the official Fraunhofer reference encoder on a university server — Fraunhofer being the German research institute that developed the MP3 and has since made hundreds of millions of dollars in licensing fees¹. SoloH modified the code to make it more usable and released it into the wild. That act, barely a footnote in mainstream accounts, ignited the era of CD ripping and peer-to-peer sharing.
This was the pivotal moment Sterne was referencing. The major US record labels were absent from this space. What began as a technical innovation, followed by a cultural hack that freed the MP3 from its institutional constraints, effectively liberated music from the existing economic enclosure. But what followed was not a new framework. The process demonetized music as an object but failed to reimagine it as anything other than the digital equivalent of smashing shop windows and grabbing whatever you could. It was a structural collapse without conceptual reconstruction. There was no serious effort to reposition music as a shared cultural resource with collective responsibilities attached — something closer to how Elinor Ostrom framed the governance of common goods³.
Without that reframing, we entered the next phase: the Napster era. In response, the record labels, through their trade group, the RIAA, launched a legal campaign that would end up suing more than 18,000 individuals⁴ for downloading or sharing their music collections across peer-to-peer networks. And when we look back on that period — of grandparents being bankrupted and families settling out of fear of financial ruin — from our current vantage point, we have to shake our heads and mutter, all that for what?
Well, there was a brief moment during the liminal stage between Napster’s demise, the launch of the iTunes Music Store in 2003, and the emergence of Pandora in 2005 when a different future seemed possible.
Economist Dean Baker proposed what he called the Artistic Freedom Voucher⁵, a publicly funded system in which every adult citizen would receive a $100 annual voucher to allocate to musicians or artists of their choosing. Funded at $20 billion per year, the plan was intended to provide a stable, decentralized source of income for cultural workers without relying on copyright enforcement or market sales. Just remove copyright entirely and use this system, plus revenue from touring and merch sales.
This was a serious attempt to reimagine how we support cultural labor rather than trying to find a way to backend existing power relations to new emerging ones. Baker’s proposal echoed Elinor Ostrom’s model for managing common goods — not through privatization or state control, but through participatory, community-based governance. In the model Baker created, music is treated as a shared cultural resource. As a responsibility we owe music, not a product to be enclosed or endlessly extracted.
But the proposal didn’t take hold. The idea was there, but the infrastructure and political imagination weren’t (we were preparing to shock and awe Iraq into freedom in 2003). As Liz Pelly notes in Mood Machine: The Rise of Spotify and the Cost of the Perfect Playlist⁶, even the people working on alternatives at the time admitted they had no perfect answer. The moment passed, and what emerged in its place wasn’t a new economic structure but a rebranding campaign.
Around 2005, I was in an office in downtown Oakland for an all-hands meeting called by our new CEO, Joe Kennedy. The company was pivoting. We were moving away from placing kiosks at the end of music aisles in Best Buy — they were stations that let you scan a CD and get recommendations for similar albums. Instead, we were going to become an online radio company. We were also changing our name from Savage Beast Technologies to Pandora.
There were maybe a dozen people in the room. I looked around to gauge the mood. It was flat. I left early and walked down Broadway toward 25th Street, where you could still park for free back then. No meters. Most of the buildings were empty. I called my dad and told him I would probably lose this job analyzing music for the Music Genome Project as we were about to compete with companies who made Rhapsody and RealPlayer, which I already used and thought we could never compete against. Also, it felt like a long shot, and I didn’t see it panning out.
Soon after, a new word colonized the digital music discourse: democratization. Pandora and soon Spotify, Apple Music, and others weren’t merely erecting digital fences around the demonetized MP3-sharing culture. Their ambition reached further. They claimed to be democratizing music, freeing it from material constraints and making it more accessible. They positioned themselves as the definitive answer to the question that had lingered since Napster: who has the right to music? Their answer came packaged in slick interfaces and frictionless streams: Everyone—the universal right to unlimited music for less than half the price of a single cassette. Revolution complete.
But what exactly was wrapped up in that word, democratization? It wasn’t a new economic model for supporting artists. It was an infrastructure designed to extract value not from music itself but from the behavioral data produced by its consumption. The songs were free. The listener paid for the right to be tracked. And the real product was everything that listening revealed.
A quick aside: I fundamentally believe Pandora was different in its mission from what came after (Spotify, Apple Music, etc.). The company really did want to direct money to artists. The founder was a former touring musician, and most of the early employees were musicians themselves in varying degrees of professionalism. I believe that materially matters. I also suspect — though I can’t definitively prove — that this is why the major labels took large equity stakes in Spotify, but not in Pandora. It’s a hunch, but it aligns with what I saw from the inside. For more on Pandora vs Spotify cultural differences, I recommend this excellent piece by Rocco Pendola a journalist who covered Pandora during this time:
If You Hate Spotify, You Would Have Loved Pandora’s Original Plans
Ultimately, capitalism and Wall Street ruin everythingmedium.com
In the streaming era, musicians quickly learned that democratization was just a new face on an old structure. What it meant was consolidation of catalogs, attention, and power. It’s easy to look back on the 1980s and 90s and resent how music had to pass through the bottleneck of the major labels. Some of the best music was only accessible on college radio or passed around on mix tapes, shared hand-to-hand like secrets. But at least back then, the scarcity was real. Music carried weight differently than today. It felt like something you had to find and care for once you had it. And, importantly, musicians and songwriters were making money, especially compared to now.
But within the logic of the streaming paradigm, music isn’t the point. The point is the data generated by people using the service and, just as critically, the data that isn’t generated when they leave.
Keeping users on-platform becomes the primary design goal. This shapes everything, including what song plays next. If a track disrupts the flow of your “lean back” listening experience, you might leave the app to learn more about the artist, or worse, skip the song, get annoyed, and close the app altogether. That kind of friction is a threat. The system learns to avoid it. Your passive listening experience must be protected at all costs, and that cost is steep. It’s paid by musicians and by the culture we’re losing in the process.
But there is more here, and this is a focus for me and a new project I am working on: this logic doesn’t just affect what we hear. It reaches backward into the songwriting process, then into the recording studio, and ultimately, the very texture of music itself.
The platform’s optimization for music that keeps people on platform is fed into the metrics it publishes which are analyzed and inform the decisions made by record labels, A & R executives, and producers, all of which create a terrifying feedback loop.
I’ve spent the past month diving into music industry podcasts, and it’s remarkable to hear label executives describe their strategy as monitoring Spotify playlists to determine what’s working — then either signing artists to match that sound or shaping songs to fit the mold. Of course, some version of this dynamic existed long before Spotify. But now we’re operating in a context where scale, efficiency, and immediate performance metrics are no longer just aesthetic pressures. What once included space for experimentation and boundary-pushing has been replaced by real-time feedback loops that create financial imperatives. These imperatives function like derivative contracts — structures that hedge against risk by narrowing the range of acceptable futures. In this model, music isn’t made to explore what’s possible; it’s made to satisfy the predictive models of those underwriting the next quarter’s returns.
In today’s music industry, access to capital to record, promote, or tour can hinge on a track’s performance within Spotify’s ecosystem. Even lending institutions focusing on the music space use these metrics to shape deal terms⁷. The music is still being made, but the surrounding system has become increasingly indifferent to meaning.
So now we have a dynamic where the vast majority of musicians earn fractions of a penny per stream on Spotify yet still exist within an ecosystem that insists they conform. They’re expected to optimize their production not for resonance or meaning but to ensure the music facilitates a lean-back experience where data can be harvested and engagement never breaks.
This ecology is what I call griddified liquidity. Music becomes flattened, standardized, and easily interchangeable. It is smoothed out for the sake of uninterrupted flow—a pure plane of immanence⁸. Any sharp edge that might provoke reflection or friction is filed down. What we’re left with is not music as a symbolic expression but as background software: endlessly playable, instantly forgettable, and perfectly suited to collecting data for those preserving jars.
Ask yourself: How many songs have you heard recently that made you want to rage against the machine? Under this operating logic, protest music has been practically eliminated—music has been transformed from music to optimizable content.
Today, we have a new entrant to this ecosystem: generative AI. Now, we no longer need the bothersome, laborious process of spending decades learning an instrument and listening to thousands of records while living in the material world. You can type a prompt into a web interface and generate a song in seconds. No calluses. No struggle. No history. No community. Just frictionless production feeding frictionless consumption for the sake of producing meaningless data.
And here we are again. Generative AI is being sold to us as democratization. The CEO of Suno.ai recently said⁹, “It’s not really enjoyable to make music now…” You see, this software will democratize your access to creation. Why should those elitist musicians be the only ones allowed to plumb the depths of the human condition? This machine, designed to replicate the artifacts of past human production, is here for you.
Here we see the ultimate logic of our modern economic system: velocity. The velocity of production — both material and digital — and the speed at which data, and the capital tied to that data, circulates is what generates billionaires. It keeps the stock charts moving up and to the right.
The biggest obstacle to velocity is material reality. So the solution is simple: build a system unmoored from physical constraint. Remove the friction of time, labor, skill, and intention — even in music. Once that happens, the upper limit on circulation disappears.
That’s how you make it rain. Not for artists, not for listeners, but for the architects of artifice. Meanwhile, the rest of us move through culture in an anxiety-ridden stupor, unable to articulate our unease — not because it isn’t real, but because we’ve lost the means to produce meaning. We’ve lost the means to listen, to relate, to just be OK with who we are.
And in this stupor, we are enacting the ultimate telos of these AI machines: to obsolesce the body. The machine’s goal is not to serve but to replace. Unlike a tool, which extends the body, a machine renders the body irrelevant.
This completes the logic of griddified liquidity. The artist isn’t erased; they’re rendered unnecessary, too risky for the velocity of data production.
In the dozens of essays I’ve written on Medium about AI-generated art, culture, music, technology, and politics, I’ve arrived at one clear conclusion: there is no way for artists — musicians, in this context — to counter this force from within. There is no winning this game. And because of that, I return to the insight of the brilliant economist Albert Hirschman: when voice no longer yields change, and loyalty is no longer reciprocated, we must choose to exit².
Any future for musicians will not be found online. It must be built offline, back in the material realm, where plumbing the depths of the human condition is laborious, slow, and filled with friction. It is risky. It is painful. It is sublime. And it is fun. Only in the material realm can the vital function of meaning-making be preserved. Only there can we hold open a space for a culture that might someday serve as a beacon when society comes looking for what it senses has been lost.
After spending all this time reflecting on the transformation into the streaming age — and now, into the generative AI age — I’ve come to believe that the only way forward is actually to build what I want to see.
Not argue for it. Not lobby for it. Build it.
So I’m working on a new and yet not-so-new model for returning music to a place where it is respected. And through that respect, to restore dignity to the musicians who bring these objects of meaning into the world. Not as data points to be harvested and sold but as carriers of culture, care, and insight that reveal to the culture of which it is born a bit about itself that is not possible through discursive and analytic means.
A little over 5 years ago, I looked around and asked myself why does everything suck so hard? That question led me to retire from music, leave my seventeen-year career at Pandora, and return to grad school. I bought and read hundreds of books — dense ones. I worked in city hall, in economic development, and various public and quasi-public institutions.
And everywhere I went, I saw the same thing I had seen in the private sector: a deep mystification about what we are doing and why we’re even doing it besides meeting some arbitrary metric on a spreadsheet.
The problem isn’t our capacity for prediction, efficiency, or innovation. The problem is that we no longer know how to make sense of the parts of our culture that cannot be counted or quantified. These aspects, often dismissed as inefficiencies or irrelevancies, are in fact the essence of the human condition. And they can only be revealed through non-discursive means¹⁰.
This is what Oscar Wilde meant when he said that art is useless. Its value lies in its inability to serve the ends of optimization, calculation, or control. It resists being folded into systems of measurement. In that resistance, it preserves a kind of insight and orientation that no algorithm can reproduce. If art is useless to the domain of production and metrics, then it is indispensable to the domain of meaning, reflection, and human truth.
If anything in this piece resonates with you, I hope you’ll take a moment to visit www.musefoundry.studio. It’s the landing page for a new kind of music label I’m building: offline, intentional, and designed to resist everything this essay critiques.
Sources
¹ Sterne, J. (2012). MP3: The meaning of a format. Duke University Press. https://doi.org/10.1215/9780822395522
² Hirschman, A. O. (1970). Exit, voice, and loyalty: Responses to decline in firms, organizations, and states. Harvard University Press.
³ Ostrom, E. (1990). Governing the commons: The evolution of institutions for collective action. Cambridge University Press.
⁴ Anderson, N. (2009, July 8). Has the RIAA sued 18,000 people… or 35,000? Ars Technica. https://arstechnica.com/tech-policy/2009/07/has-the-riaa-sued-18000-people-or-35000/
⁵ Baker, D. (2003, November 5). The artistic freedom voucher: An Internet age alternative to copyrights. Center for Economic and Policy Research. https://cepr.net/publications/the-artistic-freedom-voucher-internet-age-alternative-to-copyrights/
⁶ Pelly, L. (2025). Mood machine: The rise of Spotify and the costs of the perfect playlist. Atria/One Signal Publishers.
⁷ Sparrow, J. (Host). (2025, April 10). beatBread CEO Peter Sinclair on getting more options, insight and choices in music funding (Ep. 167) [Audio podcast episode]. In Music Ally Focus. Music Ally. https://musically.com/2025/04/10/beatbread-ceo-peter-sinclair-on-getting-more-options-insight-and-choices-in-music-funding/
⁸ Deleuze, G., & Guattari, F. (1987). A thousand plateaus: Capitalism and schizophrenia (B. Massumi, Trans.). University of Minnesota Press. https://www.upress.umn.edu/9780816614028/a-thousand-plateaus/
⁹ Stebbings, H. (Host). (2024, January 10). Mikey Shulman, CEO @Suno: The future of music, what is gonna happen?! | E1244 [Video podcast episode]. In 20VC with Harry Stebbings. YouTube.
¹⁰ Langer, S. K. (1953). Feeling and form: A theory of art developed from Philosophy in a new key. Charles Scribner’s Sons.